emo/openrc
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

librc: Complain when a real and virtual service have the same name

Browse Source
This commit is contained in:
William Hubbs 2016-01-21 15:35:55 -06:00
parent e4eacf02ca
commit 69f052b611
2 changed files with 25 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
man/openrc-run.8
View File

@ -172,9 +172,9 @@ The service will start after these services and stop before these services.
The service will start before these services and stop after these services.
.It Ic provide
The service provides this virtual service. For example, named provides dns.
Virtual services take precedence over real services, so it is highly
recommended that you do not have a real service that has the same name
as a virtual service.
Note that it is not legal to have a virtual and real service with the
same name. If you do this, you will receive an error message, and you
must rename either the real or virtual service.
.It Ic config
We should recalculate our dependencies if the listed files have changed.
.It Ic keyword

26
src/librc/librc-depend.c
View File

@ -717,14 +717,16 @@ rc_deptree_update_needed(time_t *newest, char *file)
}
librc_hidden_def(rc_deptree_update_needed)
/* This is a 6 phase operation
/* This is a 7 phase operation
Phase 1 is a shell script which loads each init script and config in turn
and echos their dependency info to stdout
Phase 2 takes that and populates a depinfo object with that data
Phase 3 adds any provided services to the depinfo object
Phase 4 scans that depinfo object and puts in backlinks
Phase 5 removes broken before dependencies
Phase 6 saves the depinfo object to disk
Phase 6 looks for duplicate services indicating a real and virtual service
with the same names
Phase 7 saves the depinfo object to disk
*/
bool
rc_deptree_update(void)
@ -733,7 +735,7 @@ rc_deptree_update(void)
RC_DEPTREE *deptree, *providers;
RC_DEPINFO *depinfo = NULL, *depinfo_np, *di;
RC_DEPTYPE *deptype = NULL, *dt_np, *dt, *provide;
RC_STRINGLIST *config, *types, *sorted, *visited;
RC_STRINGLIST *config, *dupes, *types, *sorted, *visited;
RC_STRING *s, *s2, *s2_np, *s3, *s4;
char *line = NULL;
size_t len = 0;
@ -742,6 +744,7 @@ rc_deptree_update(void)
bool retval = true;
const char *sys = rc_sys();
struct utsname uts;
int serrno;
/* Some init scripts need RC_LIBEXECDIR to source stuff
Ideally we should be setting our full env instead */
@ -996,7 +999,22 @@ rc_deptree_update(void)
}
rc_stringlist_free(types);
/* Phase 6 - save to disk
/* Phase 6 - Print errors for duplicate services */
dupes = rc_stringlist_new();
TAILQ_FOREACH(depinfo, deptree, entries) {
serrno = errno;
errno = 0;
rc_stringlist_addu(dupes,depinfo->service);
if (errno == EEXIST) {
fprintf(stderr,
"Error: %s is the name of a real and virtual service.\n",
depinfo->service);
}
errno = serrno;
}
rc_stringlist_free(dupes);
/* Phase 7 - save to disk
Now that we're purely in C, do we need to keep a shell parseable file?
I think yes as then it stays human readable
This works and should be entirely shell parseable provided that depend